WO2017203672A1

WO2017203672A1 - Item recommendation method, item recommendation program, and item recommendation apparatus

Info

Publication number: WO2017203672A1
Application number: PCT/JP2016/065646
Authority: WO
Inventors: 英俊松岡; 山本　達也; 池田　弘
Original assignee: 富士通株式会社
Priority date: 2016-05-26
Filing date: 2016-05-26
Publication date: 2017-11-30
Also published as: JP6696568B2; JPWO2017203672A1

Abstract

An item recommendation method according to one embodiment causes a computer to execute: a process for categorizing items, among a plurality of items that are compared with each other, into the same category; a process for calculating a co-occurrence probability of items combined with items that belong to a prescribed category on the basis of use history information indicating the combination of items, used by a user in the past; and a process for outputting recommended items to a category to which items selected by the user belong on the basis of the calculated co-occurrence probability.

Description

Item recommendation method, item recommendation program and item recommendation device

Embodiments of the present invention relate to an item recommendation method, an item recommendation program, and an item recommendation device.

Conventionally, with respect to the item (information) selected by the user, an item similar to the selected item is searched from a large number of items stored in a database (DB) or the like, and the searched item is recommended to the user There is an item recommendation device to do. In this item recommendation device, based on a predetermined operation performed by the user, a history of items simultaneously adopted by the user in the past is used to recommend similar items that can be simultaneously adopted for items selected by the user.

JP 2014-10715 A

However, in the above-described conventional technology, there is a problem that it is difficult to recommend an item with high accuracy when the history of items that the user adopts at the same time is small.

For example, for new items and unpopular items with few adoptions, the history of items that the user adopts at the same time decreases. Further, as the item segmentation progresses, the results of adoption of each item are dispersed, and thus history accumulation is delayed. Thus, when the history of items that the user adopts at the same time is small, it is difficult to extract similar items that can be adopted simultaneously with respect to the item selected by the user from among a number of items.

An object of the present invention is to provide an item recommendation method, an item recommendation program, and an item recommendation device that can recommend an item with high accuracy.

In the first proposal, in the item recommendation method, the computer executes processing in which items compared among a plurality of items are set to the same category. In the item recommendation method, the computer executes a process of calculating a co-occurrence probability of an item to be combined with an item belonging to a predetermined category based on the adoption history information indicating a combination of items previously adopted by the user. In the item recommendation method, the computer executes processing for outputting an item recommended for the category to which the item selected by the user belongs based on the calculated co-occurrence probability.

According to one embodiment of the present invention, items can be recommended with high accuracy.

FIG. 1 is a block diagram illustrating an example of a functional configuration of the item recommendation device according to the embodiment. FIG. 2 is a flowchart illustrating an example of processing for recording an item adoption history and a comparison group. FIG. 3 is an explanatory diagram for explaining a screen for selecting and comparing items. FIG. 4 is an explanatory diagram for explaining the employment history table. FIG. 5 is an explanatory diagram for explaining the comparison group information. FIG. 6 is a flowchart illustrating an example of processing for generating category information. FIG. 7 is a flowchart illustrating an example of a process for recommending an item. FIG. 8 is an explanatory diagram illustrating acquisition of recommended items. FIG. 9 is an explanatory diagram for explaining item recommendation. FIG. 10 is an explanatory diagram for explaining the employment history table. FIG. 11 is an explanatory diagram for explaining similar user group information. FIG. 12 is a block diagram illustrating an example of a hardware configuration of the item recommendation device according to the embodiment.

Hereinafter, an item recommendation method, an item recommendation program, and an item recommendation device according to an embodiment will be described with reference to the drawings. In the embodiment, configurations having the same functions are denoted by the same reference numerals, and redundant description is omitted. In addition, the item recommendation method, the item recommendation program, and the item recommendation device described in the following embodiments are merely examples, and do not limit the embodiments. In addition, the following embodiments may be appropriately combined within a consistent range.

FIG. 1 is a block diagram illustrating an example of a functional configuration of an item recommendation device according to an embodiment. As illustrated in FIG. 1, the item recommendation device 1 includes a storage unit 10, an input unit 20, a display unit 30, a category information generation unit 40, and a recommendation unit 50. The item recommendation device 1 is estimated to be combined with an item selected from a number of items stored in the item attribute information 11 of the storage unit 10 for an item (information) selected by the user by the input unit 20. It is a device that recommends to the user for items to be performed.

The storage unit 10 is a storage device such as a memory or HDD (Hard Disk Drive), and stores item attribute information 11, an employment history table 12, comparison group information 13, and category information 14.

Item attribute information 11 is information indicating each attribute of an item for each item. The item attributes indicate features of the item such as price, size, color, and function of the item. The category information 14 stores identification information (for example, an item ID) for identifying an item that is a target (search target) as a recommended item. Alternatively, as one of the attributes of the item attribute information 11, a category ID may be held and used as category information. At this time, the same item may hold a plurality of category IDs.

The adoption history table 12 is history information indicating a combination of items previously adopted by the user from among a plurality of items (item groups) indicated in the item attribute information 11 for each user. The adoption history table 12 is, for example, a matrix of items adopted by the user and items adopted in combination with the items, and is two-dimensional table data having the number of adoptions of the combination as a matrix element. It can also be calculated from a two-dimensional matrix that records the number of adoptions of all items for each user. The storage unit 10 stores an employment history table 12 to which identification information such as a user ID is assigned for each user.

The comparison group information 13 is information indicating items that have been compared in the past by the user among the plurality of items (item groups) indicated in the item attribute information 11 as the same category (group). For example, the comparison group information 13 stores identification information (for example, item ID) indicating an item that has been compared in the past by the user for each group ID indicating the group.

The category information 14 is information indicating an item for each category by categorizing the item attribute information 11 indicating each attribute of the item into similar items having similar attributes by a clustering method. For example, the category information 14 stores item identification information (for example, item ID) classified by clustering for each category. As one of the attributes of the item attribute information 11, the category information 14 can be substituted by holding the category ID (s).

The input unit 20 is a UI (user interface) that receives a user operation input from the input device 102 (see FIG. 12) or the like. For example, the input unit 20 receives selection of an item to be compared or adopted from the user by an operation input on a screen or the like displaying a plurality of items indicated in the item attribute information 11.

The display unit 30 performs screen display on the monitor 103 (see FIG. 12). For example, the display unit 30 reads the item attribute information 11 of the item selected by the input unit 20 and displays a list of items selected as items to be compared or adopted on the screen. Further, the display unit 30 records the history of the items that are selected as items to be compared or adopted and displayed as a list screen in the storage unit 10 as the adoption history table 12 and the comparison group information 13.

FIG. 2 is a flowchart showing an example of processing for recording an item adoption history and a comparison group. As shown in FIG. 2, when the process is started, the display unit 30 reads the item attribute information 11 and displays a plurality of items (item groups) indicated in the item attribute information 11 on a screen (S1). Here, selection of an attribute to be left as a display from among a number of attributes may be received from the input unit 20. At the same time, selection of an attribute as a sorting key may be accepted. In this case, only the attributes designated to remain as a display are displayed, and sorting is performed in the order of the sorting key and the designated attribute value. These designated items can be considered as attributes of interest to the user, and this information is used in a variation of the present invention.

Next, the input unit 20 receives selection of an item from the item group displayed on the screen from the user (S2). Next, the display unit 30 reads out the attributes (features) of the selected item from the item attribute information 11 and performs a comparison display in which the items are arranged and displayed on the screen (S3).

FIG. 3 is an explanatory diagram illustrating a screen for selecting and comparing items. As shown in FIG. 3, a plurality of items (item groups) shown in the item attribute information 11 are displayed on the selection screen G <b> 1, and the user selects items to be compared and displayed from the input unit 20. In the illustrated example, “Item (c)”, “Item (f)”, and “Item (g)” are selected as items to be compared and displayed.

The display unit 30 displays the attributes (function 1, function 2, function 3, price) of the items (“Item (c)”, “Item (f)”, and “Item (g)”) selected on the selection screen G1. Read from item attribute information 11. Next, the display unit 30 displays a list of attributes of the selected item on the comparison screen G2. Thereby, the user can compare the attribute of each item.

The user usually adopts only some items from the displayed many items. This adoption refers to an action such as purchase in electronic commerce, for example. In S2, as in S1, the input unit 20 may accept selection of an attribute to be displayed on the comparison screen G2 and an attribute to be a rearrangement key from among a number of attributes. In other words, the user may select an attribute to be confirmed at the time of comparison display or an attribute to be a sorting key. And in the display part 30, you may display the attribute selected in S2 on the comparison screen G2.

Next, the display unit 30 records the employment history table 12 and the comparison group information 13 in the storage unit 10 based on the item selected from the input unit 20 (S4).

FIG. 4 is an explanatory diagram for explaining the employment history table 12. As shown in FIG. 4, the adoption history table 12 is a matrix indicating the number of times a combination is adopted as an element of a matrix for a combination of items (a to g in the illustrated example) adopted by any user. For example, in the employment history table 12, a history (the number of times of employment) in which items are combined by values indicated by the axis item and the adopted item is shown.

In S4, the adopted item combination is recorded as a history by incrementing the number corresponding to the adopted item combination from the input unit 20 in the adopted user adoption history table 12. For example, when “Item (c)” and “Item (f)” are adopted from the comparison screen G2, the number of times corresponding to the combination of f with c as the axis, and the combination of c with f as the axis The history is recorded by incrementing the number of times.

FIG. 5 is an explanatory diagram for explaining the comparison group information 13. As shown in FIG. 5, the comparison group information 13 stores a name for identifying an item that has been compared in the past by the user for each group ID (GI0, GI1,...) Indicating a group. In S <b> 4, a group ID is issued with the combination of items selected from the input unit 20 as one group, and the name and ID of the selected item are recorded in the comparison group information 13. For example, in the example of FIG. 3, comparison group information 13 having “Item (c)”, “Item (f)”, and “Item (g)” as groups is recorded.

Referring back to FIG. 1, the category information generation unit 40 generates category information 14 by categorizing items with similar items having similar attributes by the clustering method from the item attribute information 11 indicating each attribute of the item. The process of generating the category information 14 in the category information generation unit 40 may be performed at a timing when the item attribute information 11 and the comparison group information 13 are updated, or is performed every predetermined period (for example, one month). It may be broken.

FIG. 6 is a flowchart showing an example of processing for generating the category information 14. As shown in FIG. 6, when the process is started, the category information generation unit 40 calculates the distance between each item in the feature space around each attribute (feature) of the item based on the item attribute information 11. Calculate (S10).

Next, the category information generation unit 40, based on the comparison group information 13, about the scale of the feature space based on the feature information of a plurality of items, the items belonging to the same category (group), that is, the items displayed in comparison. The distance between each item is corrected so as to reduce the distance (S11).

As a specific example, when the feature amount of an item is expressed by a multidimensional vector, the square of the distance between two items (j, k) is expressed by the following equation (1). The category information generation unit 40 calculates the distance between each item by this equation (1).

Next, the item group that is simultaneously compared, that is, the item group indicated in the comparison group information 13 is set as a comparison group, and the distance index S _gq in the q-th comparison group is set as the following expression (2).

In Expression (2), the distance index S _gq is the sum of the distances of all the two item sets in the group, but may be a distance average obtained by dividing the sum by the number of sets as another example.

Also, the distance index value S is defined as the following equation (3).

The category information generation unit 40 obtains a value of C _i (i = 1 to M) such that the distance index value S of Equation (3) is minimized under the constraint condition of Equation (4) below. That is, C _i that is converted so that the distance between items belonging to the same group is made closer to each other is obtained for the scale of the feature space based on the feature information of a plurality of items.

In order to prevent most of C _{i from} becoming 0, S _i may be added with a power of C _i . For example, the sum of squares shown in the following equation (5) may be used.

The category information generation unit 40 defines the distance by the formula (1) using the obtained C _i , so that the items belonging to the same group (the items for which comparison display has been performed) with respect to the distance between the items. The distance is corrected to be closer.

Next, the category information generation unit 40 performs clustering of items based on the corrected distances between the items (S12), and categorizes the similar items that are close to each other and have similar features. Next, the category information generation unit 40 generates category information 14 indicating similar items for each category based on the clustering result of S12 (S13). The category information generation unit 40 stores the generated category information 14 in the storage unit 10.

Returning to FIG. 1, the recommendation unit 50 receives an item selected by the user through the input unit 20 and acquires an item estimated to be combined with the item selected by the user. Next, the recommendation unit 50 outputs the acquired item as an item recommended to the user by, for example, screen display on the monitor 103 (see FIG. 12).

Specifically, the recommendation unit 50 selects items to be combined with items belonging to a predetermined category of the category information 14 based on the adoption history table 12 indicating the combination of items that the user has adopted in the past and the category information 14. Calculate the co-occurrence probability. The co-occurrence probability indicates a rate at which a plurality of items are simultaneously employed. When an item of a certain category is selected, the recommendation unit 50 also records the history of the items adopted in combination with the items that are simultaneously adopted for the category (co-occurrence probability) with the items belonging to the category. And ask. The recommendation unit 50 calculates the co-occurrence probability of items simultaneously adopted for this category (co-occurrence probability between categories) for each category of the category information 14.

Next, the recommendation unit 50 selects an item having a higher co-occurrence probability based on the co-occurrence probability of the items that are simultaneously adopted for the category to which the item selected by the user belongs than the calculated co-occurrence probability between the categories. Output as recommended items. That is, the recommendation part 50 recommends an item using the co-occurrence probability between categories instead of the inter-item co-occurrence probability in item-based collaborative filtering.

FIG. 7 is a flowchart showing an example of processing for recommending an item. As illustrated in FIG. 7, when the process is started, the recommendation unit 50 acquires a selection item selected by the user using the input unit 20 (S20).

Next, the recommendation unit 50 calculates the co-occurrence probability between categories based on the employment history table 12 and the category information 14 (S21). For example, the recommendation unit 50 uses the employment record (history) of items categorized in the category information 14 based on the employment history table 12 as the total number. Next, the recommendation unit 50 acquires, for each item categorized in the category information 14, the item as an axis item and the history of the adopted items combined with the axis item (the number of times adopted) from the adoption history table 12. Next, the recommendation unit 50 obtains the co-occurrence probability of items that are simultaneously adopted for the category from the number of times of adoption for the total number.

Next, the recommendation unit 50 acquires a recommended item having a high co-occurrence probability for the category to which the item selected by the user belongs based on the calculated co-occurrence probability between categories (S22). Next, the recommendation unit 50 outputs the acquired recommended item by screen display or the like (S23).

FIG. 8 is an explanatory diagram for explaining acquisition of recommended items. Referring to FIG. 8, tables T1 and T2 indicate that each category categorized in the category information 14 is an axis category (ka to kf), and the co-occurrence probabilities (adopted) of items (ag) that are simultaneously adopted for the category. It is a table showing the ratio). In S21, the table T1 is obtained based on the employment history table 12. Next, in S22, a recommended item having a high co-occurrence probability is acquired for the category to which the item selected by the user belongs based on the table T1. For example, when the category to which the item selected by the user belongs is “ka”, “c”, “g”, “e”, and “f” are obtained as recommended items in descending order of the ratio (co-occurrence probability). It will be.

FIG. 9 is an explanatory diagram for explaining item recommendation. In FIG. 9, it is assumed that the user “D” has selected the item “a”. Case C1 is a case in which item recommendation using inter-item co-occurrence probabilities in item-based collaborative filtering is performed. Case C2 is a case in which item recommendation is performed using the co-occurrence probability between categories instead of the co-occurrence probability between items in the item-based collaborative filtering.

As shown in FIG. 9, as the item segmentation progresses, the history of adoption of each item is dispersed, and thus the history accumulation is delayed. For this reason, in the case C1 in which the item recommendation using the inter-item co-occurrence probability is made, there is no history (adoption record) combined with the item “a”, and the recommendation is made for the user “D” who selects the item “a”. Item candidates cannot be obtained. On the other hand, in the case C2 in which item recommendation is performed using the co-occurrence probability between categories, the item “d” having a track record of adopting the category can be recommended together with the item “b” in the same category. Therefore, it is possible to recommend items with high accuracy even when the history of items that the user adopts at the same time is small.

(Modification)
Here, a modified example of the above-described embodiment will be described. In the modified example, users having similar interests about the features (attributes) of items are grouped as similar users. Then, the item recommendation is performed based on the grouped similar user employment history table 12.

Specifically, the input unit 20 of the item recommendation device 1 has, for each user, a history of attributes selected as attributes to be displayed during item group display (S1) or comparison display (S3) from among a number of attributes. Is stored in the storage unit 10.

FIG. 10 is an explanatory diagram for explaining the employment history table. As shown in FIG. 10, the input unit 20 displays the number of selections for the selected attribute (function 1, function 2,...) For each user (A, B, C...) In the adoption history table 12 a of the storage unit 10. To record.

The recommendation unit 50 groups users having similar interests regarding the selected attribute as similar users based on the recorded employment history table 12a. Specifically, the recommendation unit 50 determines the similarity of another user to a certain user based on the number of selections for the selected attribute (function 1, function 2,...) For each user in the employment history table 12a. Ask. Then, the recommendation unit 50 groups users whose similarity is equal to or greater than a predetermined threshold as similar users, and generates similar user group information indicating the grouped similar users. For example, in the example of FIG. 10, users “A” and “C” having similar selections for “function 1”, “function 3”, and “function 5” are grouped as similar users.

FIG. 11 is an explanatory diagram for explaining similar user group information. As shown in FIG. 11, the similar user group information 13a stores the names of users grouped as similar users for each group ID (GU0, GU1,...) Indicating a group.

The recommendation unit 50 obtains a similar user similar to the user who selected the item from the similar user group information 13a in S21. Then, the recommendation unit 50 calculates the co-occurrence probability between categories based on the obtained similar user employment history table 12. Thereby, in the item recommendation device 1, it is possible to recommend an item that more closely matches the user's interest based on the employment history table 12 of similar users who have similar interests about the feature (attribute) of the item.

As described above, the item recommendation device 1 sets items that have been compared from among a plurality of items based on the comparison group information 13 as the same category. In addition, the item recommendation device 1 calculates the co-occurrence probability of an item combined with an item belonging to a predetermined category based on the adoption history table 12 indicating a combination of items adopted by the user in the past. Further, the item recommendation device 1 outputs an item recommended for the category to which the item selected by the user belongs based on the calculated co-occurrence probability. Alternatively, another item with a high co-occurrence probability is output in the category to which this item belongs. In addition, in the modification of the present invention, the attribute information input for display by the user is used as an attribute of interest to find a similar user, and from among the items purchased by the similar user or items of the same category The thing with a high co-occurrence probability is output. Therefore, the item recommendation device 1 can recommend an item with high accuracy even when the history of items that the user adopts at the same time is small.

It should be noted that each component of each illustrated apparatus does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution / integration of each device is not limited to that shown in the figure, and all or a part thereof may be functionally or physically distributed or arbitrarily distributed in arbitrary units according to various loads or usage conditions. Can be integrated and configured.

The various processing functions performed in the item recommendation device 1 may be executed entirely or arbitrarily on a CPU (or a microcomputer such as an MPU or MCU (Micro Controller Unit)). In addition, various processing functions may be executed in whole or in any part on a program that is analyzed and executed by a CPU (or a microcomputer such as an MPU or MCU) or hardware based on wired logic. Needless to say, it is good. In addition, various processing functions performed in the item recommendation device 1 may be executed in cooperation by a plurality of computers by cloud computing.

Incidentally, the various processes described in the above embodiments can be realized by executing a program prepared in advance on a computer. Therefore, in the following, an example of a computer (hardware) that executes a program having the same function as the above embodiment will be described. FIG. 12 is a block diagram illustrating an example of a hardware configuration of the item recommendation device 1 according to the embodiment.

As shown in FIG. 12, the item recommendation device 1 includes a CPU 101 that executes various arithmetic processes, an input device 102 that receives data input, a monitor 103, and a speaker 104. In addition, the item recommendation device 1 includes a medium reading device 105 that reads a program and the like from a storage medium, an interface device 106 for connection to various devices, and a communication device 107 for communication connection with an external device by wire or wireless. Have. The item recommendation device 1 also includes a RAM 108 that temporarily stores various types of information and a hard disk device 109. Each unit (101 to 109) in the item recommendation device 1 is connected to the bus 110.

The hard disk device 109 stores a program 111 for executing various processes in the input unit 20, the display unit 30, the category information generation unit 40, and the recommendation unit 50 described in the above embodiment. Also, the hard disk device 109 stores various data 112 (item attribute information 11, employment history table 12, comparison group information 13, category information 14, etc.) referred to by the program 111. For example, the input device 102 receives input of operation information from an operator of the item recommendation device 1. The monitor 103 displays various screens operated by the operator, for example. The interface device 106 is connected to, for example, a printing device. The communication device 107 is connected to a communication network such as a LAN (Local Area Network), and exchanges various types of information with an external device via the communication network.

The CPU 101 reads out the program 111 stored in the hard disk device 109, develops it in the RAM 108, and executes it to perform various processes. Note that the program 111 may not be stored in the hard disk device 109. For example, the program 111 stored in a storage medium readable by the item recommendation device 1 may be read and executed. The storage medium readable by the item recommendation device 1 corresponds to, for example, a portable recording medium such as a CD-ROM or a DVD disk, a USB (Universal Serial Bus) memory, a semiconductor memory such as a flash memory, a hard disk drive, or the like. Alternatively, the program 111 may be stored in a device connected to a public line, the Internet, a LAN, or the like, and the item recommendation device 1 may read and execute the program 111 therefrom.

DESCRIPTION OF SYMBOLS 1 ... Item recommendation apparatus 10 ... Memory | storage part 11 ... Item attribute information 12, 12a ... Employment history table 13 ... Comparison group information 13a ... Similar user group information 14 ... Category information 20 ... Input part 30 ... Display part 40 ... Category information generation part 50 ... Recommended unit 101 ... CPU
102 ... Input device 103 ... Monitor 104 ... Speaker 105 ... Media reader 106 ... Interface device 107 ... Communication device 108 ... RAM
109 ... Hard disk device 110 ... Bus 111 ... Program 112 ... Various data C1, C2 ... Case G1 ... Selection screen G2 ... Comparison screen T1, T2 ... Table

Claims

Items that have been compared among multiple items are in the same category,
Calculate the co-occurrence probability of an item combined with an item belonging to a predetermined category based on the employment history information indicating a combination of items that the user has adopted in the past,
An item recommendation method, wherein a computer executes a process of outputting an item recommended for a category to which an item selected by the user belongs based on the calculated co-occurrence probability.
Receiving selection of an item to be compared from the plurality of items;
The computer further executes a process of displaying the feature information of the selected item side by side,
The item recommendation method according to claim 1, wherein in the process of making the same category, the selected item is made the same category.
The computer further executes a process of performing a conversion to reduce the distance between items belonging to the same category for the scale of the feature space based on the feature information of the plurality of items,
The item recommendation method according to claim 1, wherein the calculating process calculates the co-occurrence probability based on the converted feature space.
The accepting process accepts selection of at least one feature information from a plurality of feature information,
The item recommendation method according to claim 1, wherein the displaying process displays the selected feature information side by side with respect to the selected item.
The computer further executes a process in which a user having similar feature information selected from the plurality of feature information is a similar user,
The item recommendation method according to claim 4, wherein the calculating process calculates the co-occurrence probability based on adoption history information indicating a combination of items adopted by the similar user in the past.
Items that have been compared among multiple items are in the same category,
Calculate the co-occurrence probability of an item combined with an item belonging to a predetermined category based on the employment history information indicating a combination of items that the user has adopted in the past,
An item recommendation program that causes a computer to execute a process of outputting an item recommended for a category to which an item selected by the user belongs based on the calculated co-occurrence probability.
Receiving selection of an item to be compared from the plurality of items;
Causing the computer to further execute a process of displaying the feature information of the selected item side by side;
The item recommendation program according to claim 6, wherein the same category processing sets the selected items to the same category.
For the scale of the feature space based on the feature information of the plurality of items, the computer is further caused to perform a process of performing a conversion to reduce the distance between items belonging to the same category,
The item recommendation program according to claim 6 or 7, wherein the calculating process calculates the co-occurrence probability based on the converted feature space.
The accepting process accepts selection of at least one feature information from a plurality of feature information,
The item recommendation program according to claim 6, wherein the displaying process displays the selected feature information side by side with respect to the selected item.
Causing the computer to further execute a process in which a user having similar feature information selected from the plurality of feature information is a similar user;
The item recommendation program according to claim 9, wherein the calculating process calculates the co-occurrence probability based on adoption history information indicating a combination of items previously adopted by the similar user.
Processor
Items that have been compared among multiple items are in the same category,
Calculate the co-occurrence probability of an item combined with an item belonging to a predetermined category based on the employment history information indicating a combination of items that the user has adopted in the past,
An item recommendation device that executes a process of outputting an item recommended for a category to which the item selected by the user belongs based on the calculated co-occurrence probability.
Receiving selection of an item to be compared from the plurality of items;
The processor further executes a process of displaying the feature information of the selected item side by side,
The item recommendation device according to claim 11, wherein in the process of making the same category, the selected item is made the same category.
The processor further executes a process of performing a conversion to reduce the distance between items belonging to the same category for the scale of the feature space based on the feature information of the plurality of items,
The item recommendation device according to claim 11 or 12, wherein the calculating process calculates the co-occurrence probability based on the converted feature space.
The accepting process accepts selection of at least one feature information from a plurality of feature information,
The item recommendation device according to claim 11, wherein the displaying process displays the selected feature information side by side with respect to the selected item.
The processor further executes a process in which a user having similar feature information selected from the plurality of feature information is a similar user,
The item recommendation device according to claim 14, wherein the calculating process calculates the co-occurrence probability based on adoption history information indicating a combination of items that the similar user has adopted in the past.